The present invention generally relates to the formation of materials useful for electronic applications, in particular for photovoltaic solar cells and, more specifically, for forming films of such materials from particulate precursors.
Manufacturers of electronic devices, in particular photoelectronic devices and more specifically photovoltaic devices, are increasingly employing non-elemental materials such as III-V, II-VI and I-III-VI semiconductors and alloys, mixtures and layered structures of such materials. The constituent layers in such photoelectronic devices are typically fabricated using vapor phase deposition processes such as vacuum evaporation, sputtering and chemical vapor deposition. Vapor phase processes are useful for small-scale research and for high-precision processing of high-value, small-area devices such as integrated circuits. Vapor phase processes have yielded solar cells with high sunlight-to-electricity conversion efficiencies; but it is difficult to deposit uniform films on large areas using coincident vapor phase processes in which the constituent elements are co-deposited, hence coincident vapor phase processes are typically costly to scale up to large device sizes with the control and through-put required for commercial production.
Various researchers have explored sequential vapor phase processes. CuInSe.sub.2 films are typically formed by sequentially depositing solid layers of Cu and In elemental metals and subsequently reacting the Cu--In composite layers with a source of Se to form CuInSe.sub.2 By solid layer we mean a substantially solid layer of material with minimal included void space. Multi-step, sequential processes substitute separate sequential deposition of constituents in place of co-deposition of constituents, with the intent of mitigating materials interactions that typically complicate co-deposition processes. However, this separation into sequential depositions of different constituents can introduce new complications; for example, vapor phase deposited indium tends to de-wet, forming localized islands, and solid layers of Cu--In alloys can segregate into In-rich and Cu-rich phases when heated, with the result that extreme care is required to maintain the desired planar Cu--In layer structure and lateral composition uniformity during the early stages of subsequent reactions to form CuInSe.sub.2.
Various researchers have explored techniques for stabilizing metal precursor layers such as Cu--In for CuInSe.sub.2 against de-wetting and phase segregation. For example, thin layers of a chalcogenide metal such as tellurium can be deposited on the substrate prior to the deposition of Cu and In in order to form telluride compounds that mitigate indium de-wetting and phase segregation, and solid layers of binary chalcogenide compounds such as Cu.sub.2 Se and In2Se.sub.3 can be deposited and subsequently interdiffused to form ternary CuInSe.sub.2. Such processes use chalcogenide compounds to stabilize the primary non-chalcogen constituent metals Cu and In against phase segregation during deposition and subsequent processing.
Other researchers have explored vapor phase processing of oxide-containing phase-stabilized precursors and mixtures of such precursors with elemental metals and non-oxide chalcogenides. For example, chalcogenide solid films can be formed by depositing a metal oxide solid film and annealing the oxide film in a gas or vapor containing a metal chalcogen such as S, Se, Te or a mixture thereof. Such processes can also utilize layers of single-phase oxide particulates, such as Cu.sub.2 In.sub.2 O.sub.5 particles or such as Cu.sub.2 O particles mixed with In.sub.2 O.sub.3 particles, and can also utilize solid layers of a mixture of metal and oxide, selenide or sulfide compound constituents. While mitigating certain complications of sequential processes, these phase-stabilized precursor improvements leave unresolved the inherent complexities of achieving the constituent layer uniformity necessary to achieve high-quality semiconductor films using vapor phase processes.
Various researchers have explored alternatives to vapor phase processes for fabricating semiconductor materials for various photoelectronic applications. Electrodeposition can be used to sequentially deposit solid layers of constituent metals such as Cu and In that are subsequently reacted with chalcogenide metals such as Se to form compound semiconductor films. The chalcogenide metal can be embedded in the electroplated solid metal layers by adding Se particles to the electroplating bath so as to incorporate Se particles into one or more of the metallic layers. Additional Se can be added by screen printing a solid Se layer on the Cu--In--Se precursor layers, or by annealing the electrodeposited layers in Se vapor. While avoiding some of the disadvantages of vapor phase processes, such electrodeposition processes are plagued with the challenges of uniform high-rate electrodeposition on large-area substrates and introduce electrodeposition-specific complications such as metal-contaminated waste treatment, recovery and disposal.
Other researchers have explored alternatives to both vapor phase and electrodeposition processes. For example, spray pyrolysis techniques can be used to deposit metal oxide solid layers, and the oxide layers can subsequently be annealed in chalcogenide metal vapor to convert the oxide layers to chalcogenide films. Spray pyrolysis is convenient for depositing multi-component oxides on large areas; but the materials use efficiencies of spray pyrolysis processes are generally low, hence manufacturing materials costs are generally high.
Alternatively, one can form precursor layers by screen printing a paste of particles or by painting a substrate with a slurry of particles. For example, one can form a Cu--In--Se powder, prepare a paste from the powder, screen print layers of the paste, and anneal the layers to form CuInSe.sub.2 films. Cu--In--Se powders prepared by ball milling or grinding reportedly yield median particle diameters of 1.5 .mu.m and larger. Median powder particle diameter determines minimum pinhole-free CuInSe.sub.2 layer thickness; particle diameters of 1.5-2 .mu.m typically limit CuInSe.sub.2 film thickness to 5 .mu.m or greater. Such film thickness are a factor of 5-10 thicker than necessary to absorb incident sunlight, and result in high manufacturing materials costs. Researchers preparing CuInSe.sub.2 films by screen printing and sintering CuInSe.sub.2 -based pastes report taking particular measures to avoid the formation of indium oxides deleterious to the CuInSe.sub.2 film properties. Screen-printed films are typically much thicker than required to absorb sunlight sufficiently. Screen printing and related film formation processes are unlikely to be economic unless effective strategies for forming powders with smaller particle diameters and for processing the powders to achieve good film quality are developed.
Various researchers have investigated small particles with median particle diameters of 100 nm and less as a pathway to preparing thin-film materials. Nanoparticles of a wide range of oxides (e.g. ZnO, SnO.sub.2, WO.sub.3, etc.) and chalcogenides (e.g. CdS, CdTe, etc.) have been reported, and thin films have been formed from nanoparticles by a variety of techniques. Such small particles can be deposited as particulate layers by a variety of processes including, for example, electrophoresis of colloidal suspensions and slurry spraying. CuInSe.sub.2 films have been prepared by spraying slurries of mixtures of single-phase, binary selenide nanoparticles such as Cu.sub.x Se and In.sub.x Se, but film quality and device performance are poor due to insufficient interparticle diffusion. This implies that small median chalcogenide particle diameters alone do not provide improved particle-based thin film properties.
The use of an effective flux is known to be particularly important for promoting particle coalescence and grain growth in particle-based thin films. CdCl.sub.2 works well as a flux with Cd-based chalcogenide materials such as CdTe and Cd(Se,Te). A comparable flux has not been reported for CuInSe.sub.2 and related alloys. Se, CuCl, InCl.sub.3 and Cu-Se compounds have been evaluated as fluxes for screen-printed CuInSe.sub.2 layers, and non-negligible fluxing reportedly occurs only at relatively high temperatures in Cu-rich material when a liquid Cu.sub.2 Se is present. Thus effective fluxing of CuInSe.sub.2 is possible only under conditions that result in Cu-rich CuInSe.sub.2 films unsuitable for solar cells, or in complex multi-step processes such as continuous co-evaporation in which the growing CuInSe.sub.2 film is temporarily made Cu-rich to effect transient fluxing via liquid Cu.sub.2 Se before differentially adding In and Se to achieve an In-rich final film composition. Results are also poor when paste mixtures of Cu and In elemental powders are screen printed and overcoated with screen-printed Se powder paste.
Summarizing the prior art, coincident vapor phase processes are difficult to control on large areas; sequential vapor phase processes simplify some of the complexities, but composition uniformity and precursor phase segregation can be severe problems. Metal oxides are useful as phase-stabilized precursors, but uniform, high-rate, vapor phase deposition of oxide solid films is difficult, and spray pyrolysis of oxide solid films is inefficient due to poor materials use efficiencies. Particulate-based processes such as screen printing can have high materials efficiencies, but such processes generally work well only when efficient fluxing processes are available. A clear need exists for phase-stabilized precursors that can be easily converted to thin-film, photovoltaic materials. A need also exists for an easily controlled process for forming such precursors. It would be especially advantageous to provide a process using particulate precursors to fabricate thin films with well-controlled compositions on large areas, as well as materials and processing techniques for creating high quality, thin film products without contamination from residual flux materials.